Video Game (Electronic Arts) Global Sales Analysis¶
Main Goal¶
- The goal of this project is to show Exploratory Data Analysis and Predictive Analysis of Electronic Arts Global Sales
Data Description¶
Dataset can be downloaded from Video Game Sales
This dataset contains a list of video games with sales greater than 100,000 copies. It was generated by a scrape of vgchartz.com
The dataset contains 11 columns. The names and data types are as follows:
- Rank - Ranking of overall sales, integer
- Name - The games name, object
- Platform - Platform of the games release (i.e. PC,PS4, Nintendo, etc.), object
- Year - Year of the game's release, float
- Genre - Genre of the game ,object
- Publisher - Publisher of the game
- NA_Sales - Sales in North America (in millions)
- EU_Sales - Sales in Europe (in millions)
- JP_Sales - Sales in Japan (in millions)
- Other_Sales - Sales in the rest of the world (in millions)
- Global_Sales - Total worldwide sales.
Data Processing & Wrangling¶
In [1]:
# Load important packages
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import missingno as msno
import plotly.graph_objs as go
from plotly.offline import init_notebook_mode,iplot
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import cross_val_score, cross_val_predict
In [2]:
# Load Dataset
df = pd.read_csv("dataset/vgsales.csv", encoding="ISO-8859-1", low_memory=False)
df.head()
Out[2]:
In [3]:
df.shape
Out[3]:
Handling missing values¶
In [4]:
msno.matrix(df)
Out[4]:
In [5]:
missing_data = df.isnull()
for column in missing_data.columns.values.tolist():
print(column)
print (missing_data[column].value_counts())
print("")
- Missing values are in Publisher and Year columns. Let's drop those values.
In [6]:
df.dropna(inplace=True)
In [7]:
df.isnull().sum()
Out[7]:
In [8]:
df.shape
Out[8]:
In [9]:
df.dtypes
Out[9]:
In [10]:
# Coverting "Year" column to datetime
df['Year'] = pd.to_datetime(df['Year'], format='%Y')
In [11]:
df['Year'].dtypes
Out[11]:
In [12]:
df.head()
Out[12]:
In [13]:
df['year'] = df['Year'].dt.year
In [14]:
df = df.drop(['Year'], axis=1)
In [15]:
df.head()
Out[15]:
Checking the number of unique values¶
In [16]:
df_num_unique = df[['Genre', 'Publisher', 'Platform', 'Name']]
print(df_num_unique.nunique())
Checking some basic statistics from Sales¶
In [17]:
df_sales = df[['NA_Sales', 'EU_Sales', 'JP_Sales', 'Other_Sales', 'Global_Sales']]
df_sales.describe()
Out[17]:
Exploratory Data Analysis¶
Genre¶
In [18]:
genre_count = pd.DataFrame(df['Genre'].value_counts())
genre_percent = pd.DataFrame(df['Genre'].value_counts(normalize = True))
genre_df = genre_count.merge(genre_percent, left_index = True, right_index = True)
genre_df['Genres'] = genre_df.index
genre_df.columns = ['Count', 'Percent','Genres']
genre_df = genre_df.reset_index().drop('index', axis = 1)
genre_df.head(10)
Out[18]:
In [19]:
df["Genre"].value_counts().plot.barh()
Out[19]:
Publisher¶
In [20]:
publisher_count = pd.DataFrame(df['Publisher'].value_counts())
publisher_percent = pd.DataFrame(df['Publisher'].value_counts(normalize = True))
publisher_df = publisher_count.merge(publisher_percent, left_index = True, right_index = True)
publisher_df['Publisher'] = publisher_df.index
publisher_df.columns = ['Count', 'Percent','Publisher']
publisher_df = publisher_df.reset_index().drop('index', axis = 1)
publisher_df.head(10)
Out[20]:
- As shown above, Electronic Arts has published majority of the games in this dataset; therefore, we will mainly focus on Electronic Arts Sales.
Electronic Arts North America Sales by Genres¶
In [21]:
ea_na = df[(df["Publisher"] == "Electronic Arts")]
plt.figure(figsize=(20, 6))
sns.lineplot(data=ea_na, x="year", y="NA_Sales", hue="Genre")
Out[21]:
Electronic Arts Sales in Europe (in millions) by Genres¶
In [22]:
ea_eu = df[(df["Publisher"] == "Electronic Arts")]
plt.figure(figsize=(20, 6))
sns.lineplot(data=ea_eu, x="year", y="EU_Sales", hue="Genre")
Out[22]:
Electronic Arts Sales in Japan (in millions) by Genres¶
In [23]:
ea_na = df[(df["Publisher"] == "Electronic Arts")]
plt.figure(figsize=(10, 6))
sns.lineplot(data=ea_na, x="year", y="JP_Sales", hue="Genre")
Out[23]:
Electronic Arts Sales in the rest of the world (in millions)by Genres¶
In [24]:
ea_rest = df[(df["Publisher"] == "Electronic Arts")]
plt.figure(figsize=(20, 6))
sns.lineplot(data=ea_rest, x="year", y="Other_Sales", hue="Genre")
Out[24]:
Electronic Arts Sales in Total worldwide sales.by Genres¶
In [25]:
ea_global = df[(df["Publisher"] == "Electronic Arts")]
plt.figure(figsize=(20, 6))
sns.lineplot(data=ea_global, x="year", y="Global_Sales", hue="Genre")
Out[25]:
Platform¶
In [26]:
platform_count = pd.DataFrame(df['Platform'].value_counts())
platform_percent = pd.DataFrame(df['Platform'].value_counts(normalize = True))
platform_df = platform_count.merge(platform_percent, left_index = True, right_index = True)
platform_df['Platform'] = platform_df.index
platform_df.columns = ['Count', 'Percent','Platform']
platform_df = platform_df.reset_index().drop('index', axis = 1)
platform_df.head(10)
Out[26]:
In [27]:
df_EA = df[(df["Publisher"] == "Electronic Arts")]
In [28]:
xaction=df_EA[df_EA.Genre=="Action"]
xsports=df_EA[df_EA.Genre=="Sports"]
xmisc=df_EA[df_EA.Genre=="Misc"]
xrole=df_EA[df_EA.Genre=="Role-Playing"]
xshooter=df_EA[df_EA.Genre=="Shooter"]
xadventure=df_EA[df_EA.Genre=="Adventure"]
xrace=df_EA[df_EA.Genre=="Racing"]
xplatform=df_EA[df_EA.Genre=="Platform"]
xsimulation=df_EA[df_EA.Genre=="Simulation"]
xfight=df_EA[df_EA.Genre=="Fighting"]
xstrategy=df_EA[df_EA.Genre=="Strategy"]
xpuzzle=df_EA[df_EA.Genre=="Puzzle"]
In [29]:
trace1 = go.Histogram(
x=xaction.Platform,
opacity=0.75,
name = "Action",
marker=dict(color='rgb(165,0,38)'))
trace2 = go.Histogram(
x=xsports.Platform,
opacity=0.75,
name = "Sports",
marker=dict(color='rgb(215,48,39)'))
trace3 = go.Histogram(
x=xmisc.Platform,
opacity=0.75,
name = "Misc",
marker=dict(color='rgb(244,109,67)'))
trace4 = go.Histogram(
x=xrole.Platform,
opacity=0.75,
name = "Role Playing",
marker=dict(color='rgb(253,174,97)'))
trace5 = go.Histogram(
x=xshooter.Platform,
opacity=0.75,
name = "Shooter",
marker=dict(color='rgb(254,224,144)'))
trace6 = go.Histogram(
x=xadventure.Platform,
opacity=0.75,
name = "Adventure",
marker=dict(color='rgb(170,253,87)'))
trace7 = go.Histogram(
x=xrace.Platform,
opacity=0.75,
name = "Racing",
marker=dict(color='rgb(171,217,233)'))
trace8 = go.Histogram(
x=xplatform.Platform,
opacity=0.75,
name = "Platform",
marker=dict(color='rgb(116,173,209)'))
trace9 = go.Histogram(
x=xsimulation.Platform,
opacity=0.75,
name = "Simulation",
marker=dict(color='rgb(69,117,180)'))
trace10 = go.Histogram(
x=xfight.Platform,
opacity=0.75,
name = "Fighting",
marker=dict(color='rgb(49,54,149)'))
trace11 = go.Histogram(
x=xstrategy.Platform,
opacity=0.75,
name = "Strategy",
marker=dict(color="rgb(10,77,131)"))
trace12 = go.Histogram(
x=xpuzzle.Platform,
opacity=0.75,
name = "Puzzle",
marker=dict(color='rgb(1,15,139)'))
trace = [trace1, trace2,trace3,trace4,trace5,trace6,trace7,trace8,trace9,trace10,trace11,trace12]
layout = go.Layout(barmode='stack',
title='Electronic Arts Genre Counts According to Platform',
xaxis=dict(title='Platform'),
yaxis=dict( title='Count'),
paper_bgcolor='beige',
plot_bgcolor='beige'
)
fig = go.Figure(data=trace, layout=layout)
iplot(fig)
- As shown from above graph, we see that Electronic Arts has made a number of sports games in Playstation and XBOX platform.
Now let's check the yearly global sales of Electronic Arts by Genres using heatmap
In [30]:
pivot = df_EA.pivot_table(index="Genre",
values="Global_Sales",
columns="year",
aggfunc="sum")
In [31]:
plt.figure(figsize=(30,10))
sns.heatmap(pivot, cmap="Blues", annot=True, fmt=".0f")
Out[31]:
- As expected, Sports has the most global sales especially from 2004 to 2010.
Now Let's check their most renowned game FIFA by Platform
In [32]:
df_fifa = df[df['Name'].str.match('FIFA')]
In [33]:
df_fifa['Name'].unique()
Out[33]:
In [34]:
pivot_fifa = df_fifa.pivot_table(index="Platform",
values="Global_Sales",
columns="Name",
aggfunc="sum")
In [35]:
plt.figure(figsize=(35,12))
sns.heatmap(pivot_fifa, cmap="Blues", annot=True, fmt=".0f")
Out[35]:
- As shown above, FIFA12, FIFA 13, and FIFA14 are really popular in PS3 platform and FIFA15 and FIFA16 are really popular in PS4 platform.
Let's check the time series plot of FIFA Global Sales
In [36]:
fifa_global_sales_yearly = df_fifa.groupby(by=['year'], as_index=False)['Global_Sales'].sum()
plt.figure(figsize=(20,5))
plt.plot(fifa_global_sales_yearly.year, fifa_global_sales_yearly.Global_Sales)
plt.show()
Predictive Analysis¶
In [37]:
# Checking Correlation
# df_EA = df[(df["Publisher"] == "Electronic Arts")]
sns.heatmap(df_EA.corr(),cbar=True,annot=True)
Out[37]:
In [38]:
#df_EA.head()
Decision Tree Regression¶
In [39]:
features = ['NA_Sales','EU_Sales','JP_Sales','Other_Sales']#our features
X = df_EA[features]
y = df_EA.Global_Sales #target
X_train , X_test , y_train , y_test = train_test_split(X,y,test_size=0.25,random_state=1)
In [40]:
DTmodel = DecisionTreeRegressor(random_state=1)
DTmodel.fit(X_train,y_train)
Out[40]:
In [41]:
predictions = DTmodel.predict(X_test)
In [42]:
mae = mean_absolute_error(predictions, y_test)
mse = mean_squared_error(predictions, y_test)
print('Mean Absolute Error: '+str(mae))
print('Mean Squared Error: '+str(mse))
In [43]:
pred_EA_df = pd.DataFrame({'Actual': y_test, 'Predicted': predictions})
pred_EA_df.head()
Out[43]:
In [44]:
score = DTmodel.score(X_test, y_test)
In [45]:
score
Out[45]:
Random Forest Regression¶
In [46]:
features = ['NA_Sales','EU_Sales','JP_Sales','Other_Sales']#our features
X = df_EA[features]
y = df_EA.Global_Sales #target
X_train , X_test , y_train , y_test = train_test_split(X,y,test_size=0.25,random_state=1)
In [47]:
RFmodel = RandomForestRegressor(n_estimators=10, random_state = 0)
RFmodel.fit(X_train, y_train)
Out[47]:
In [48]:
predictions = RFmodel.predict(X_test)
In [49]:
mae = mean_absolute_error(predictions, y_test)
mse = mean_squared_error(predictions, y_test)
print('Mean Absolute Error: '+str(mae))
print('Mean Squared Error: '+str(mse))
In [50]:
pred_EA_df = pd.DataFrame({'Actual': y_test, 'Predicted': predictions})
pred_EA_df.head()
Out[50]:
In [51]:
score = RFmodel.score(X_test, y_test)
score
Out[51]:
Ridge Regression¶
In [52]:
from sklearn.preprocessing import LabelEncoder
cat_var = ['Platform', 'Genre', 'Name']
num_var = ['NA_Sales','EU_Sales','JP_Sales','Other_Sales','Global_Sales']
LE = LabelEncoder()
label_encode = pd.DataFrame(columns=['Platform', 'Genre', 'Name','Global_Sales'])
# for var in cat_var:
# col = df_EA[var]
# new_col = LE.fit_transform(col)
# label_encode[var] = new_col
# for var in num_var:
# label_encode[var] = df_EA[var].values
# label_encode.head()
In [53]:
def encoding(cat_var, num_var):
for var in cat_var:
col = df_EA[var]
new_col = LE.fit_transform(col)
label_encode[var] = new_col
for var in num_var:
label_encode[var] = df_EA[var].values
return(label_encode)
In [54]:
ridge_df = encoding(cat_var, num_var)
In [55]:
X = ridge_df[['Platform', 'Genre', 'NA_Sales', 'EU_Sales', 'JP_Sales', 'Other_Sales']]
y = ridge_df['Global_Sales']
X_train , X_test , y_train , y_test = train_test_split(X,y,test_size=0.25,random_state=1)
In [56]:
from sklearn.linear_model import Ridge
ridge = Ridge()
ridgeModel = ridge.fit(X_train, y_train)
predictions = ridgeModel.predict(X_test)
In [57]:
mae = mean_absolute_error(predictions, y_test)
mse = mean_squared_error(predictions, y_test)
print('Mean Absolute Error: '+str(mae))
print('Mean Squared Error: '+str(mse))
In [58]:
pred_EA_df = pd.DataFrame({'Actual': y_test, 'Predicted': predictions})
pred_EA_df.head()
Out[58]:
In [59]:
score = ridgeModel.score(X_test, y_test)
score
Out[59]:
In [ ]: